Gender Classification based on Local Binary Pattern and K-Nearest Neighbour

Gender recognition from face image is a research area in the field of pattern recognition that determines gender based on biometric features of the face. Research has shown that the disparity between facial masculinity and femininity can be utilized to improve performances of face recognition applications in biometrics, human‐computer interaction, and surveillance and computer vision. There is need to develop a gender classification system for human‐computer interaction that recognizes the gender of an individual through the input face image. Hence, this study developed a facial gender recognition system using K‐Nearest Neighbour (KNN). Facial images from the FERET database were obtained from the internet. The image database is made up of several images of male and female faces which have been categorized based on gender. These databases contain images of different poses for each individual. Local Binary pattern (LBP) was used to extract features of the images and the images were classified using K‐Nearest Neighbour algorithm. The system was implemented using Matrix Laboratory 8.1 (MATLAB 2015a). The classification results showed that highest accuracy of 92% was achieved. The system could be adopted in classifying face images into male or female which is required in security control system or any other related systems. Keyword— Feature extraction, Gender, Human Computer interaction, K‐Nearest Neighbour algorithm Local Binary pattern (LBP) face classification, training and testing